公开(公告)号：KR100449486B1

公开(公告)日：2004-09-22

申请号：KR1020010088362

申请日：2001-12-29

Applicant: 한국전자통신연구원

Inventor： 김두식 ,  김호연 ,  임길택 ,  송재관 ,  남윤석 ,  김혜규

IPC: G06K9/20

CPC classification number: G06K9/00463

Abstract: In a document recognition system, a document structure analysis unit extracts a character image region from an input document image. A character string extraction unit extracts a character string image from the character image region. A character extraction unit extracts an individual character image from the character string image expressed in vertical lines by vertical line adjacency graphs by changing a pixel representation of the extracted character string image into a vertical line representation thereof. A character recognition unit recognizes each character in the individual character image and converting the recognized character into a corresponding character code.

Abstract translation: 在文件识别系统中，文件结构分析单元从输入文件图像中提取字符图像区域。 字符串提取单元从字符图像区域中提取字符串图像。 字符提取单元通过将所提取的字符串图像的像素表示改变为其垂直线表示，通过垂直线邻接图从垂直线表示的字符串图像中提取单独的字符图像。 字符识别单元识别单个字符图像中的每个字符并将识别的字符转换为对应的字符代码。

公开(公告)号：KR3003606040000S

公开(公告)日：2004-09-07

申请号：KR3020030031185

申请日：2003-11-05

Applicant: 한국전자통신연구원

Designer： 김혜규 ,  김인수 ,  이석

公开(公告)号：KR3003606030000S

公开(公告)日：2004-09-07

申请号：KR3020030031184

申请日：2003-11-05

Applicant: 한국전자통신연구원

Designer： 김혜규 ,  김인수 ,  이석

公开(公告)号：KR1020040057175A

公开(公告)日：2004-07-02

申请号：KR1020020083632

申请日：2002-12-24

Applicant: 한국전자통신연구원

Inventor： 박문성 ,  박상은 ,  정윤수 ,  김혜규 ,  이웅희 ,  정동석

IPC: G06K7/10

Abstract: PURPOSE: A system for fast reading an ROI(Region Of Interest) of a barcode and a controlling method thereof are provided to recognize the material information fast by recognizing a sort of the barcode, and calculating the area information of the material in order to minimize a time for searching the barcode area. CONSTITUTION: A postal matter image storage(125) stores a postal matter image, the candidate barcode region images, and the barcode sort information. A barcode characteristic value comparator(121) minimizes a wrongly detected region from the candidate barcode regions printed on the postal matter. A barcode ROI extractor(122) extracts the sort and the exterior coordinates by generating a common test line for the barcode, and generates a central axis line easy to read the barcode by the outer coordinates in case that only the linear barcode is detected. A barcode information extractor(123) generates the location and thickness information of a bar thickness and spatial value after scanning the coordinates region of the postal matter image. A barcode information decoder(124) generates a read result of the barcode.

Abstract translation: 目的：提供一种用于快速读取条形码的ROI（感兴趣区域）的系统及其控制方法，以便通过识别条形码的类别来快速识别材料信息，并且计算材料的区域信息以便最小化 搜索条形码区域的时间。 规定：邮政图像存储部（125）存储邮件图像，候选条形码区域图像和条形码排序信息。 条形码特征值比较器（121）使打印在邮件上的候选条形码区域的错误检测区域最小化。 条形码ROI提取器（122）通过生成条形码的公共测试线来提取排序和外部坐标，并且在仅检测到线性条形码的情况下，通过外部坐标生成容易读取条形码的中心轴线。 条形码信息提取器（123）在扫描邮件图像的坐标区域之后生成条厚度和空间值的位置和厚度信息。 条形码信息解码器（124）生成条形码的读取结果。

公开(公告)号：KR1020040055936A

公开(公告)日：2004-06-30

申请号：KR1020020082416

申请日：2002-12-23

Applicant: 한국전자통신연구원

Inventor： 김호연 ,  김두식 ,  임길택 ,  이상호 ,  남윤석 ,  김혜규

IPC: G06K9/00

Abstract: PURPOSE: A device and a method for recognizing a character based on a characteristic of an outline are provided to develop the characteristic having a structural characteristic and enhance a character recognition performance by extracting the characteristic based on the outline while a merit of a statistical method is kept. CONSTITUTION: An image input part(14) obtains a character image from an inputted image. A character extractor(16) extracts the outline information matched with the character image. A characteristic extractor(18) traces the outline extracted from the character extractor and extracts the structural characteristic from each point. A character recognizer(20) finds out a linear or non-linear mesh by using a black pixel density of the obtained image, generates a characteristic vector by accumulation the structural characteristic depending on the mesh, and matches/recognizes the character by using the characteristic vector.

Abstract translation: 目的：提供一种基于轮廓特征来识别角色的装置和方法，以通过提取基于轮廓的特征来提高具有结构特征的特征并提高字符识别性能，而统计方法的优点是 保持。 构成：图像输入部（14）从输入图像获得文字图像。 字符提取器（16）提取与字符图像匹配的轮廓信息。 特征提取器（18）跟踪从字符提取器提取的轮廓，并从每个点提取结构特征。 字符识别器（20）通过使用所获得的图像的黑色像素密度来找出线性或非线性网格，通过根据网格累积结构特征来产生特征向量，并且通过使用特征来匹配/识别字符 向量。

公开(公告)号：KR1020040055505A

公开(公告)日：2004-06-26

申请号：KR1020020082198

申请日：2002-12-21

Applicant: 한국전자통신연구원

Inventor： 최지영 ,  이태한 ,  김진석 ,  김혜규

IPC: G06Q10/08

Abstract: PURPOSE: A system for planning the main transportation of a hub-and-spoke material distribution network and a plan generating method thereof are provided to perform an optimal transportation plan for the main transportation of each vehicle by equipping with a device for receiving/storing a control command of a user and generating the transportation plan of the hub-and-spoke distribution network. CONSTITUTION: A user interface(10) receives the control command of the user. A temporary database(18) temporarily stores the control command. A planning engine(14) generates/evaluates/improves a main transportation path of each vehicle depending on the control command. A master database(20) stores the generated transportation paths, the standard information of the transportation path, and the evaluating/improved data. A control engine(12) generates/evaluates/improves the main transportation path of each vehicle through the planning engine according to the control command, and stores the data in the master database.

Abstract translation: 目的：提供一种用于规划轮辐和轮辐物料分配网络的主要运输的系统及其计划生成方法，以通过装备用于接收/存储轮胎的装置来执行每辆车的主要运输的最佳运输计划 控制用户的命令并生成中心辐射分发网络的运输计划。 构成：用户界面（10）接收用户的控制命令。 临时数据库（18）临时存储控制命令。 计划引擎（14）根据控制命令生成/评估/改进每个车辆的主要运输路径。 主数据库（20）存储生成的传送路径，传送路径的标准信息和评价/改进数据。 控制引擎（12）根据控制指令通过规划引擎生成/评价/改进各车辆的主要运送路径，并将数据存储在主数据库中。

公开(公告)号：KR1020030062774A

公开(公告)日：2003-07-28

申请号：KR1020020003159

申请日：2002-01-19

Applicant: 한국전자통신연구원

Inventor： 정윤수 ,  박문성 ,  이용준 ,  김인수 ,  김진석 ,  김혜규 ,  원종운

IPC: G01B11/04

Abstract: PURPOSE: A method and a system for measuring dimensions of a three-dimensional object are provided to simultaneously measure dimensions of a hexahedral object and objects having predetermined shapes. CONSTITUTION: An image taking section(100) takes an image of a three-dimensional object to be measured. A focus area detecting section(102) detects a focus area by processing an image taken by the image taking section(100). A minimum circumscribed rectangular pattern detecting section(104) detects a minimum circumscribed rectangular pattern of an object. A height information detecting section(106) detects height of the object. A dimension calculating section(108) calculates dimensions of the object based on height information of the height information detecting section(106) and the minimum circumscribed rectangular pattern detected by the minimum circumscribed rectangular pattern detecting section(104).

Abstract translation: 目的：提供一种用于测量三维物体的尺寸的方法和系统，以同时测量六面体物体和具有预定形状的物体的尺寸。 构成：摄像部（100）拍摄要测量的三维物体的图像。 聚焦区域检测部分（102）通过处理由图像拍摄部分（100）拍摄的图像来检测对焦区域。 最小外接矩形图案检测部（104）检测物体的最小外接矩形图案。 高度信息检测部（106）检测物体的高度。 尺寸计算部（108）基于高度信息检测部（106）的高度信息和由最小外接矩形图案检测部（104）检测出的最小外接矩形图案来计算对象的尺寸。

公开(公告)号：KR1020030059499A

公开(公告)日：2003-07-10

申请号：KR1020010088362

申请日：2001-12-29

Applicant: 한국전자통신연구원

Inventor： 김두식 ,  김호연 ,  임길택 ,  송재관 ,  남윤석 ,  김혜규

IPC: G06K9/20

CPC classification number: G06K9/00463

Abstract: PURPOSE: A system and a method for recognizing a document using a vertical line adjacency graph are provided to presume an image division position by using the vertical line adjacency graph and to correctly extract a partial image based on the presumed image division position. CONSTITUTION: A document structure analyzer(104) extracts a text image area by separating the text image area and a picture image area from a document image inputted from a scanner(102). A text string extractor(106) extracts a text string image for the text image area. A text extractor(108) extracts an individual text image for the text string image. The test string extractor(106) more correctly discriminates the division position of the individual text image from the text string image by presuming the image division point as using the vertical line adjacency graph, and by extracting the individual text image from the text string image based on the presumed image division point.

Abstract translation: 目的：提供使用垂直线相邻图来识别文档的系统和方法，以通过使用垂直线相邻图来假定图像分割位置，并且基于假定的图像分割位置正确地提取部分图像。 构成：文档结构分析器（104）通过从扫描仪（102）输入的文档图像中分离文本图像区域和图像图像区域来提取文本图像区域。 文本字符串提取器（106）提取文本图像区域的文本字符串图像。 文本提取器（108）提取文本串图像的单个文本图像。 通过将图像分割点设为使用垂直线相邻图，并且通过从文本串图像中提取单个文本图像，测试字符串提取器（106）更正确地区分单个文本图像与文本串图像的分割位置 在假设的图像分割点。

公开(公告)号：KR100388061B1

公开(公告)日：2003-06-18

申请号：KR1020010003223

申请日：2001-01-19

Applicant: 한국전자통신연구원

Inventor： 박문성 ,  송재관 ,  황재각 ,  남윤석 ,  김혜규 ,  박치항

IPC: G06K1/12

CPC classification number: G06K7/1473 ,  B07C3/10 ,  G06K7/14 ,  G06K19/063 ,  G07B2017/00588 ,  G07B2017/00717

Abstract: A 4-state bar code printing and reading system for use in physical distribution-related services such as mail pieces, receptacles, reception and management forms or the like, and a method for controlling the system are disclosed. The 4-state bar code printing system comprises a bar code information acquiring section for acquiring, storing and determining a mail piece sorting information; an information recorded density enhancing section for evaluating a value of a character, by using of a compression method depending upon a type of the information and digit, which are determined by the bar code information acquiring section; an input information encoding section for arranging the values of the bars produced by the information recorded density enhancing section in order of values of bars of the 4-state bar code, and dividing the values into groups each consisting of 3 bars; an error correcting codeword producing section for producing an error correcting codeword based on the result of the information encoding section the input value of an error correcting level; and a print frame producing section for arranging the error correcting codeword produced from the error correcting codeword producing section and the data produced from the information recorded density improving section in order of a start bar, a data, an error correcting codeword, and a stop bar, to print a 4-state bar code print font on a mail piece.

Abstract translation: 公开了用于诸如邮件，容器，接收和管理形式等物理分配相关服务的4状态条形码打印和读取系统，以及用于控制该系统的方法。 该4状态条形码打印系统包括用于获取，存储和确定邮件分拣信息的条形码信息获取部分; 信息记录密度增强部分，用于通过根据由条形码信息获取部分确定的信息和数字的类型的压缩方法来评估字符的值; 输入信息编码部分，用于按照4状态条形码的值的顺序排列由信息记录密度增强部分产生的条的值，并将这些值分成每个包括3个条的组; 纠错码字产生部分，用于根据信息编码部分的结果产生纠错码字;纠错电平的输入值; 以及打印帧产生部分，用于按照开始条，数据，纠错码字和停止条的顺序排列从错误校正码字产生部分产生的错误校正码字和从信息记录密度改善部分产生的数据 ，在邮件上打印4状态条码打印字体。

公开(公告)号：KR1020020051601A

公开(公告)日：2002-06-29

申请号：KR1020000080997

申请日：2000-12-23

Applicant: 한국전자통신연구원

Inventor： 이용준 ,  김혜규 ,  류근호

IPC: G06F17/30

Abstract: PURPOSE: A system and a method for searching temporal relation rules in consideration of time intervals are provided to efficiently search temporal relation rules from data equipped with time intervals by providing a new temporal data mining technique and a system structure capable of implementing the technique as software. CONSTITUTION: A sorting unit sorts an input database in customer IDs and transaction times. A frequent case sort generating unit finds a set of frequent cases at the minimum support, which is more then the support designated by a user, from the database arrayed in the array unit. A generalization unit converts the database arrayed into a case sequence with time intervals according to a time unit and a standard point designated. A set of uniformed cases with the minimum and more is found. After that, the standard point is calculated and arrayed by summarizing and generalizing the case sequence. Therefore, a time interval database is found.

Abstract translation: 目的：提供一种考虑时间间隔搜索时间关系规则的系统和方法，通过提供新的时间数据挖掘技术和能够将该技术实现为软件的系统结构来有效地从配备有时间间隔的数据中搜索时间关系规则 。 规定：排序单元将客户ID和交易时间中的输入数据库排序。 频繁案例分类生成单元从阵列单元中排列的数据库中找到最小支持的一组频繁情况，即多个由用户指定的支持。 泛化单元根据时间单位和指定的标准点将数据库排列成具有时间间隔的案例序列。 发现一套具有最小和更多的制服案例。 之后，通过总结和概括病例序列来计算和排列标准点。 因此，找到时间间隔数据库。